Underwater image dehaze using scene depth estimation with adaptive color correction
Authors:
- 1 June 2017
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-5
Abstract
Underwater images suffer from poor visibility due to color casts and light scattering that caused by physical properties existing in underwater environments. Degraded underwater images lead to a low accuracy rate of underwater object detection and recognition. To solve this problem, a novel underwater image enhancement method which combines adaptive color correction and image dehazing based on atmospheric scattering model is proposed in this paper. As the most important component of dehazing model, a transmission map is derived from the color corrected image. Considering the exponential relationship between the transmission map and scene depth map, transmission map estimation will be naturally formulated into a scene depth map estimation problem. To predict scene depth map, a Convolutional Neural Network (CNN) is employed on image patches extracted from the color corrected image. The experimental results show that the proposed strategy improves the quality of underwater images efficiently and arrives at good results in underwater objects detection and recognition.Keywords
This publication has 15 references indexed in Scilit:
- Predicting Depth, Surface Normals and Semantic Labels with a Common Multi-scale Convolutional Architecture2015
- Enhancement of low quality underwater image through integrated global and local contrast correctionApplied Soft Computing, 2015
- A Patch-Structure Representation Method for Quality Assessment of Contrast Changed ImagesIEEE Signal Processing Letters, 2015
- Review on underwater image restoration and enhancement algorithms2015
- Deep convolutional neural fields for depth estimation from a single image2015
- Underwater image segmentation via dark channel prior and multiscale hierarchical decomposition2015
- Mixture contrast limited adaptive histogram equalization for underwater image enhancement2013
- Single Image Haze Removal Using Dark Channel PriorIEEE Transactions on Pattern Analysis and Machine Intelligence, 2010
- Make3D: Learning 3D Scene Structure from a Single Still ImageIEEE Transactions on Pattern Analysis and Machine Intelligence, 2008
- Distinctive Image Features from Scale-Invariant KeypointsInternational Journal of Computer Vision, 2004